{"ID":2850057,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.22693","arxiv_id":"2510.22693","title":"VADTree: Explainable Training-Free Video Anomaly Detection via Hierarchical Granularity-Aware Tree","abstract":"Video anomaly detection (VAD) focuses on identifying anomalies in videos. Supervised methods demand substantial in-domain training data and fail to deliver clear explanations for anomalies. In contrast, training-free methods leverage the knowledge reserves and language interactivity of large pre-trained models to detect anomalies. However, the current fixed-length temporal window sampling approaches struggle to accurately capture anomalies with varying temporal spans. Therefore, we propose VADTree that utilizes a Hierarchical Granularityaware Tree (HGTree) structure for flexible sampling in VAD. VADTree leverages the knowledge embedded in a pre-trained Generic Event Boundary Detection (GEBD) model to characterize potential anomaly event boundaries. Specifically, VADTree decomposes the video into generic event nodes based on boundary confidence, and performs adaptive coarse-fine hierarchical structuring and redundancy removal to construct the HGTree. Then, the multi-dimensional priors are injected into the visual language models (VLMs) to enhance the node-wise anomaly perception, and anomaly reasoning for generic event nodes is achieved via large language models (LLMs). Finally, an inter-cluster node correlation method is used to integrate the multi-granularity anomaly scores. Extensive experiments on three challenging datasets demonstrate that VADTree achieves state-of-the-art performance in training-free settings while drastically reducing the number of sampled video segments. The code will be available at https://github.com/wenlongli10/VADTree.","short_abstract":"Video anomaly detection (VAD) focuses on identifying anomalies in videos. Supervised methods demand substantial in-domain training data and fail to deliver clear explanations for anomalies. In contrast, training-free methods leverage the knowledge reserves and language interactivity of large pre-trained models to detec...","url_abs":"https://arxiv.org/abs/2510.22693","url_pdf":"https://arxiv.org/pdf/2510.22693v3","authors":"[\"Wenlong Li\",\"Yifei Xu\",\"Yuan Rao\",\"Zhenhua Wang\",\"Shuiguang Deng\"]","published":"2025-10-26T14:36:15Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false,"code_links":[{"ID":607762,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2850057,"paper_url":"https://arxiv.org/abs/2510.22693","paper_title":"VADTree: Explainable Training-Free Video Anomaly Detection via Hierarchical Granularity-Aware Tree","repo_url":"https://github.com/wenlongli10/VADTree","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}